mcd-demo

What is mcd-demo?

mcd-demo is a project for testing the creation of simple MCP (Model Context Protocol) servers and integrating them with a LangChain agent. It is aimed at developers experimenting with MCP server implementations and LangChain agent integration.

How to use mcd-demo?

Set up a Python virtual environment, install dependencies from requirements.txt, and configure environment variables for Azure OpenAI API key and endpoint. Then start the three MCP servers (weather_server.py, math_server.py, telemetry_server.py) in the background, optionally running the math server via Docker, and finally launch the agent with python agent.py.

Key features of mcd-demo

• Demonstrates creation of simple MCP servers
• Integrates MCP servers with a LangChain agent
• Includes weather, math, and telemetry server examples
• Supports running math server in a Docker container
• Uses SSE transport for MCP server communication

Use cases of mcd-demo

• Learning how to build and run MCP servers
• Testing LangChain agent integration with multiple MCP services
• Experimenting with MCP server deployment (native or Docker)
• Prototyping multi-server MCP architectures

FAQ from mcd-demo

What MCP servers are included?

Three servers: weather_server.py, math_server.py, and telemetry_server.py.

Can the math server be run in Docker?

Yes, you can build a Docker image with make build and run it with make run-local.

What environment variables are required?

AZURE_OPENAI_API_KEY and AZURE_OPENAI_ENDPOINT are required. Optionally, MCP_MATH_URI can be set to configure the math server URI.

How do I stop the MCP servers?

Use pkill -9 -f <server_script.py> on each server, e.g., pkill -9 -f weather_server.py.

Is there a .env file supported?

Yes, variables can be set in a .env file in the project root directory.